Across the Indo-Gangetic Plain of Pakistan, India, Nepal and Bangladesh, wheat is sown as the monsoon heat recedes in November and harvested before the return of summer in March and April. The crop therefore lives almost entirely inside the cool season, and so do its three rust diseases: stripe or yellow rust caused by Puccinia striiformis f. sp. tritici, leaf or brown rust caused by P. triticina, and stem or black rust caused by P. graminis f. sp. tritici. Each fungus has its own range of favourable temperature, and the few degrees that separate a cold January from a mild one can decide which of them builds up, in which districts, and for how many weeks (Roelfs et al., 1992).

Rust populations are not uniform. Each consists of races, also called pathotypes, that differ in which host resistance genes they can overcome, a relationship formalised in the gene-for-gene concept (Flor, 1971). Wheat breeding in South Asia has leaned heavily on race-specific genes, and the working life of a cultivar depends on how long the races able to attack it stay rare. The history of the region is one of repeated boom and bust, in which a widely grown cultivar selects the race that defeats it (Wellings, 2011).

Winters in the region are becoming warmer, and heat late in the season is already recognised as a constraint on the crop itself (Lobell et al., 2012). Whether that warming is also changing the race composition of rust populations is a narrower question, and a harder one. This article reviews what the published literature can and cannot say about it. It reports no new survey or experiment.

This is a narrative review of the published literature. It draws on the classical accounts of rust epidemiology on the subcontinent, on descriptions of how races are identified and surveyed, on experimental work comparing the temperature responses of old and recent isolates, and on general reviews of climate change and plant disease. No formal search protocol was applied and no claim of completeness is made. Stripe rust receives most attention because it is the rust for which adaptation to temperature has been studied most directly. Much of that experimental work was done in North America and Europe, and its relevance to South Asia is argued here, not assumed.

Two features of the regional system matter for any argument about warming: the thermal niches of the three pathogens, and the fact that none of them survives the summer in the plains.

A race is identified by inoculating seedlings of a set of differential wheat lines, each carrying a known resistance gene, and scoring the infection types under standard conditions. The approach goes back to the stem rust differentials of Stakman et al. (1962) and, for stripe rust, to the nomenclature proposed by Johnson et al. (1972); the resistance genes themselves are catalogued by McIntosh et al. (1995). National programmes in South Asia maintain their own differential sets and analyse samples sent in from trap nurseries and farmers' fields every season. A published account of the stripe rust pathotypes found in India over a decade illustrates the kind of record these surveys produce (Prashar et al., 2007).

Three properties of such data limit what they can show about climate. Samples are not random: they come from the cultivars and districts where rust was noticed. Infection types are scored at a fixed seedling temperature, so a race label says nothing about how the isolate performs in a warm or a cold field. And races are defined by virulence alone, although traits such as latent period and spore production are what temperature acts on most directly. Genotyping has begun to fill the gap. Population studies with molecular markers have traced the worldwide lineages of the stripe rust fungus (Ali et al., 2014), and sequencing of infected leaves taken straight from the field can assign samples to lineages without culturing them (Hubbard et al., 2015).

The literature suggests four routes by which warmer winters could change race composition. Table 1 sets them out with the kind of evidence available for each, and Figure 1 places them between the climate signal and the survey record. They are discussed in turn.

The dominant cause of race change is the host. When one cultivar or one resistance gene covers a large share of the wheat area, the race able to attack it increases, and this has been seen so often that it serves as the default explanation for any shift. A climate effect has to be detected against that background. General reviews of climate and plant disease have warned for many years that responses measured in growth chambers do not translate simply into field outcomes, because host, pathogen and management all adjust at once (Garrett et al., 2006; Chakraborty and Newton, 2011).

For South Asia specifically, the published record supports three statements. Stripe rust lineages with greater tolerance of warmth exist and have spread widely. The regional epidemiology offers several points at which winter temperature could change the make-up of the founding population. And no study reviewed here has separated a temperature effect on race frequencies from the effect of cultivar turnover in the same years. The proposition that warming winters are shifting race composition in the region is thus reasonable, mechanistically grounded and not yet demonstrated.

Attribution would need data that survey programmes do not routinely collect. The most direct step is to phenotype isolates for temperature response as well as virulence: latent period, lesion growth and spore production at a cool and a warm regime, following the design of Milus et al. (2009). Applied to isolates kept in national collections from earlier decades, this would show whether thermal traits have moved, and whether they moved with particular races. A second step is to tie every survey sample to a location, a date, a cultivar and the nearest weather record, so that race frequencies can be analysed with cultivar area as a covariate.

Genotyping should accompany pathotyping as a matter of course, since lineage identity is what links a South Asian isolate to the experimentally characterised strains elsewhere (Hubbard et al., 2015). Monitoring of off-season wheat and barberry in the hills would test the survival pathway directly. On the host side, the practical response does not depend on resolving attribution: resistance that combines several partial genes, and that holds at the higher temperatures now expected late in the season, is a sound objective whatever the cause of race change (Singh et al., 2005).

Race composition in the wheat rusts of South Asia is shaped first by the cultivars grown and by races arriving from elsewhere. Warmer winters add a further influence whose mechanisms are well understood in outline: selection among pathogen lineages for performance in the warmth, changes in survival and epidemic timing, and altered expression of host resistance. Experimental evidence for the first is strong in other regions. Evidence that any of them has measurably altered race frequencies in South Asia is still lacking, mainly because surveys record virulence and not thermal traits. Adding temperature-response phenotyping, genotyping and weather records to the existing survey system would turn a plausible hypothesis into a testable one.

To be declared by the author(s) before publication.

To be declared by the author(s) before publication.